Summary: Depression and anxiety are prevalent, debilitating, and poorly understood disorders. RDoC charts the nature of these conditions across multiple units, but the domains are based on expert consensus, permitting bias and missed opportunities. Moreover, little is known about how adversity affects RDoC constructs and contributes to psychopathology. Thus, there is a critical need to rigorously evaluate RDoC domains in developmental samples from diverse backgrounds at increased risk for exposure to adversity and later psychopathology. We will use data-driven analytics to design, apply and validate multilevel-multimodal models of Threat and Reward constructs in an existing longitudinal cohort at risk for psychopathology. To predict internalizing symptoms, we will identify biotypes cross-sectionally and examine the longitudinal plasticity of RDoC-informed biotypes. Harsh social-ecological conditions will be deeply assessed and used to forecast the onset/intensification of internalizing symptoms at multiple units. We will assess 1,000 young adults from The Fragile Families and Child Wellbeing Study (FFCWS), an ongoing study of children born to predominantly low-income families. Attributes of the FFCWS are: 1) children were assessed at birth, 1, 3, 5, 9, 15 years; 2) the sample is representative of people born in cities and, thus, unlike almost all other neuroimaging research, findings are generalizable; 3) Although a full range of incomes are represented, there is substantial enrichment for low-income and African-American families, populations often under-represented in research; and 4) participants are entering early adulthood, a period of heightened risk for psychopathology. We will assess Threat and Reward at four units of analysis: symptoms, task-based behaviors, brain, and genomics and link these units to exposure to adversity. The central hypothesis is that the RDoC Threat and Reward constructs will each cluster across individuals and units, are distinct from each other, and have specific socio- ecologic predictors. We will examine multisource/multimodal data structure in 1000 participants cross- sectionally and 213 participants longitudinally. Our transdisciplinary team of experts positions us well to elucidate the structure of the Threat and Reward constructs and map risk for internalizing biotypes. We will dramatically expand our established protocol to harmonize, aggregate, cross-sectionally and longitudinally analyze, cluster, and visualize the high-dimensional datasets. Using data-driven validation approaches at four units of analysis, we will examine three aims: Aim 1 will examine RDoC Threat construct cross-sectionally, developmentally, and ecologically. Aim 2 will test RDoC Reward construct cross-sectionally, developmentally, and ecologically. Aim 3 will assess the degree to which Threat and Reward dissociate cross-sectionally, developmentally and ecologically. By deeply phenotyping a large cohort enriched for low income and African American participants, we will determine the validity of Threat/Reward and findings will generalize to a population underrepresented in research and disproportionately affected by adversity.